@article{ART003399311},
author={Chae-Jin LIM and Jung Soo Han and Hyun-Seob Lee},
title={Design and Validation of Jamo-Tag-Based Annotation for Small-Scale, Low-Resolution Images of Handwritten Hangul Syllables},
journal={Journal of Internet of Things and Convergence},
issn={2466-0078},
year={2026},
volume={12},
number={5},
pages={6}
TY - JOUR
AU - Chae-Jin LIM
AU - Jung Soo Han
AU - Hyun-Seob Lee
TI - Design and Validation of Jamo-Tag-Based Annotation for Small-Scale, Low-Resolution Images of Handwritten Hangul Syllables
JO - Journal of Internet of Things and Convergence
PY - 2026
VL - 12
IS - 5
PB - The Korea Internet of Things Society
SP - 6
EP -
SN - 2466-0078
AB - Handwritten Hangul (Korean Characters) data are often limited by insufficient samples for individual syllable classes, substantial class imbalance, and low resolution in older materials. To address these issues, a jamo-tag-based structured annotation method, termed Jamo-Structured, is proposed, in which jamo compositional information (initial, medial, and final components of a Hangul syllable) is incorporated into complete-syllable labels and training target distributions. Experiments were conducted on 3,455 low-resolution handwritten complete-syllable images from 534 classes extracted from the 1955 –1956 installments of the newspaper comic strip Oldman GoBau. In Jamo-Structured, 0.95 is assigned to the target class and the remaining 0.05 to structural neighbors. Under the same CNN (Convolutional Neural Network) architecture and data conditions, Jamo-Structured was compared with Hard Label, Uniform Label Smoothing, and Matched-Random. Across three repeated Train-Test splits, Jamo-Structured achieved a Macro-F1 of 89.55±1.82% and a Balanced Accuracy of 92.94±1.51%, the highest average performance among the comparison conditions. Its Macro-F1 exceeded those of Hard Label, Uniform Label Smoothing, and Matched-Random by 4.99, 3.59, and 9.68 percentage points, respectively. These results indicate that Jamo-Structured target distributions can improve low-resolution handwritten Hangul complete-syllable classification under small-scale, class-imbalanced conditions.
KW - Handwritten Hangul recognition;complete-syllable classification;structured annotation; Hangul jamo tags;target distribution;label smoothing
DO -
UR -
ER -
Chae-Jin LIM, Jung Soo Han and Hyun-Seob Lee. (2026). Design and Validation of Jamo-Tag-Based Annotation for Small-Scale, Low-Resolution Images of Handwritten Hangul Syllables. Journal of Internet of Things and Convergence, 12(5), 6.
Chae-Jin LIM, Jung Soo Han and Hyun-Seob Lee. 2026, "Design and Validation of Jamo-Tag-Based Annotation for Small-Scale, Low-Resolution Images of Handwritten Hangul Syllables", Journal of Internet of Things and Convergence, vol.12, no.5 6.
Chae-Jin LIM, Jung Soo Han, Hyun-Seob Lee "Design and Validation of Jamo-Tag-Based Annotation for Small-Scale, Low-Resolution Images of Handwritten Hangul Syllables" Journal of Internet of Things and Convergence 12.5 6 (2026) : 6.
Chae-Jin LIM, Jung Soo Han, Hyun-Seob Lee. Design and Validation of Jamo-Tag-Based Annotation for Small-Scale, Low-Resolution Images of Handwritten Hangul Syllables. 2026; 12(5), 6.
Chae-Jin LIM, Jung Soo Han and Hyun-Seob Lee. "Design and Validation of Jamo-Tag-Based Annotation for Small-Scale, Low-Resolution Images of Handwritten Hangul Syllables" Journal of Internet of Things and Convergence 12, no.5 (2026) : 6.
Chae-Jin LIM; Jung Soo Han; Hyun-Seob Lee. Design and Validation of Jamo-Tag-Based Annotation for Small-Scale, Low-Resolution Images of Handwritten Hangul Syllables. Journal of Internet of Things and Convergence, 12(5), 6.
Chae-Jin LIM; Jung Soo Han; Hyun-Seob Lee. Design and Validation of Jamo-Tag-Based Annotation for Small-Scale, Low-Resolution Images of Handwritten Hangul Syllables. Journal of Internet of Things and Convergence. 2026; 12(5) 6.
Chae-Jin LIM, Jung Soo Han, Hyun-Seob Lee. Design and Validation of Jamo-Tag-Based Annotation for Small-Scale, Low-Resolution Images of Handwritten Hangul Syllables. 2026; 12(5), 6.
Chae-Jin LIM, Jung Soo Han and Hyun-Seob Lee. "Design and Validation of Jamo-Tag-Based Annotation for Small-Scale, Low-Resolution Images of Handwritten Hangul Syllables" Journal of Internet of Things and Convergence 12, no.5 (2026) : 6.
